Resilient intrusion detection system for adversarial attacks on Low-Rate DDoS
Ayat Droos, Qasem S. Abu Al-Haija · Research Square · 2023
Abstract With the increase of attacks and illegal activities on networks, there has become an urgent need for systems that detect these activities and attacks and protect networks, such as intrusion detection systems that have an important role in protecting computer networks. Still, an adversarial attack has recently appeared against these systems, adversely affecting their performance. This paper proposes a framework that uses a Generative Adversarial Network (GAN) to generate adversarial instances to deceive IDS. The framework consists of two phases: building IDS models using two datasets and generating adversarial examples using GAN. Through careful evaluation, the GAN model achieves an impressive 99.9% success rate in generating samples that successfully evade intrusion detection systems, which sheds light on the potential vulnerabilities of IDSs for adversarial attacks.